1. Field of the Invention
The present invention relates to methods for image processing using motion vectors in imaging devices.
2. Description of the Related Art
Conventional methods have been proposed, in which motion vectors between frames or fields of moving images that are captured are detected for reducing the blurring of images, where the blurring can be caused by, for example, by hand-shake that occurs at the time of shooting the images. The conventional systems also determine whether a subject is a moving subject, and can also perform image compression.
Blurring of an image is reduced by, for example, shifting an image area to be output of the image obtained from an image pickup element or moving a part of an optical system in the direction perpendicular to the optical axis, on the basis of the result of detecting the motion vectors.
Unfortunately, for example, when the viewing angle is changed by optical zooming or an image is magnified or reduced by electronic zooming, the motion vector caused by the zooming operation may be erroneously detected as a motion vector caused by shaking of the device. This is because a motion vector is caused by a change in the viewing angle or an image.
Thus, a method for reducing blurring of an image is discussed in Japanese Patent Laid-Open No. 7-203280 (especially, paragraphs 19 to 27 and FIGS. 6 to 10). In this method, the position, or the size of a detection area for detecting motion vectors is changed according to the state of zooming, and motion vectors are detected for a plurality of representative points in the detection area. When these motion vectors satisfy predetermined conditions, it is assumed that these motion vectors are not those caused by zooming but those caused by shaking of the device, and the subsequent blurring of the image is reduced. Thus, even when an imaging device has a zooming function, blurring of an image can be reduced using motion vectors.
However, even in the method discussed in Japanese Patent Laid-Open No. 7-203280, detection of motion vectors caused by shaking of the device while zooming is performed and reduction of the blurring of an image on the basis of the result of the detection cannot be performed. That is to say, when shaking of the device occurs while zooming is performed, it is difficult to extract only the motion vectors caused by the shaking of the device from the detected motion vectors. This is because only the vector sum of the motion vector caused by zooming and the motion vector caused by shaking of the device is detected when the shaking of the device occurs while zooming is performed.
Moreover, recently, the speed of a zooming operation has been improved. In this situation, when motion vectors caused by zooming extend beyond a detection area for detecting motion vectors, the motion vectors cannot be detected.